In March 2007, BELDEN, a company renowned for its cable and cabling systems, acquired HIRSCHMANN, a leading global provider of industrial networking solutions, for approximately $260 million in cash. This move surprised the industry: a company that previously only operated at the physical layer acquired a current leader in industrial networking. This should prompt us to refocus on the lowest and most easily overlooked layer of ISO standards—the physical layer. Because the physical layer of ISO's Industrial Ethernet defines the mechanical, electrical, functional, and procedural characteristics between communication endpoints for activation, maintenance, and deactivation, it provides a physical medium for data transmission for the data link layer and path layer protocols. This layer is often overlooked due to its relative simplicity. Therefore, this article will focus on the Industrial Ethernet physical layer, discussing specific technologies adopted to meet the requirements of industrial environments and systems, as well as future changes to the physical layer of industrial networks. In fact, Ethernet technology has been around for nearly thirty years, during which time it has undergone numerous improvements and optimizations based on its initial half-duplex, media-sharing technology. Especially since 1994, Ethernet, due to its adoption of 10/100BASE-T technology, has distinguished itself from the network operating system wars of ATM, FDDI, and TOKEN RING, and has become the preferred choice in the LAN (Cal Area Network) and WAN (Wide Area Network) markets. Its widespread commercial application has also made it a high-bandwidth, high-speed network technology. Even after industrial applications, Ethernet can still be considered a very effective network, provided that a set of effective technologies accumulated over the years of Ethernet's evolution are selected and combined. The physical layer of industrial networks mainly includes industrial network cabling, network topology, and front-end devices with NICs (Network Controllers). Details familiar to industrial automation engineers, such as cable media and connectors in the cabling system, will not be elaborated upon here. This article will only discuss the working mechanisms of industrial network cabling, network topology, and front-end devices with NICs. I. About Industrial Network Cabling The physical layer cabling system should include both wired and wireless network cabling systems. Here, we will only describe the wired industrial network cabling system for now. The biggest difference between industrial Ethernet physical layer cabling and commercial networks lies in the cabling objects and the cabling environment. There is considerable research on the industrial network cabling environment, so it will not be elaborated upon here. Cabling Objects Commercial networks have a well-defined cabling system in 1991 by AT&T (the predecessor of Avaya). A commercial cabling system is a network cabling system designed for buildings or building complexes. It connects voice and data communication equipment, switching equipment, and other information management systems to each other, as well as to external communication networks. It includes all cables and associated cabling components between the connection points on the building to the external network or telephone exchange line and the voice or data terminals in the work area. Systems serving buildings or building complexes built according to a campus layout do not include telephone exchange network facilities, nor do they include switching equipment connected to the cabling system, such as dedicated PBXs, packet switching equipment, or the terminal equipment itself. A cabling system consists of different series of components, including: transmission media, line management hardware, connectors, sockets, plugs, adapters, transmission electronics, electrical protection devices, and support hardware. These components are used to build various subsystems, each with its specific purpose, which are not only easy to implement but also allow for a smooth transition to enhanced cabling technologies as communication needs change. A well-designed cabling system offers a degree of independence for the equipment it serves and can interconnect many different communication devices, such as data terminals, analog and digital telephones, personal computers and mainframes, and public system equipment. To understand and predict the future of the physical layer of industrial Ethernet, it is essential to study Ethernet physical layer systems; to study Ethernet physical layer cabling systems, one should begin with AVAYA's SYSTIMAX PDS system. SYSTIMAX PDS System Building structured cabling systems can be divided into six subsystems: * Work area subsystem * Horizontal cabling subsystem * Backbone subsystem * Equipment room subsystem * Management subsystem * Building complex subsystem. The various needs of communication and data processing systems determine the required subsystems. Theoretically, large communication systems may require integrating all of the above subsystems using copper and fiber optic media components. The work area subsystem consists of the cabling (or cords) connecting the terminal equipment to the information outlets. It includes mounting cords, connectors, and extension cords for the connections, bridging between the terminal equipment and I/O. While some transmission electronics may be needed for terminal equipment and I/O connections, these are not part of the work area subsystem. For example, a limited-distance modem provides the necessary signal conversion for compatibility and extended transmission distances between the terminal and other devices. Limited-distance modems do not require internal protection circuitry, but most modems do. The horizontal cabling subsystem is part of the overall cabling system, extending the backbone subsystem lines to the user's work area. The difference between the horizontal cabling subsystem and the backbone subsystem is that the horizontal cabling subsystem is always located on a single floor and terminates at information outlets. In existing buildings, the subsystem consists of 25-pair cables; however, SYSTIMAX PDS limits this to 4-pair UTP (unshielded twisted pair), which supports most modern communication equipment. Fiber optic cables can be used for certain broadband applications. Starting from the information outlet in the user work area, the horizontal cabling subsystem terminates at a crossover point; or in a small communication system, it interconnects at any of the following locations: satellite junction box, trunk junction box, or equipment room. In the equipment room, when terminal equipment is located on the same floor, the horizontal cabling subsystem terminates at a cabling crossover point. On the upper floors, it terminates at a crossover point in the trunk junction box or satellite junction box. Management Subsystem The management subsystem consists of crossovers, interconnects, and I/O. Management points provide connectivity to other subsystems. Crossovers and interconnects allow you to locate or relocate communication lines to different parts of the building for easier management of communication lines. I/O points are located in the user work area and other rooms, allowing for easy plugging and unplugging of mobile terminal equipment. When using jumpers or inserts, crossovers allow you to connect a communication line terminated on one end of a unit to a line terminated on the other end of the unit. A jumper cable is a short, single conductor that connects the ends of two wires at a crossover point; a plug cable contains several wires, each with a connector at one end. Plug cables provide a simple method for rearranging cabling and do not require the special tools used with jumpers. Interconnects achieve the same purpose as crossovers but do not use jumpers or plug cables, only wires with plugs, sockets, and adapters. Both interconnects and crossovers are used with fiber optic cables. Fiber optic crossovers require the use of fiber optic plug cables—short fiber optic cables with connectors at both ends. Plug cables can be used at various crossover points depending on the cabling arrangement and the need to manage communication lines to accommodate changes in the location of terminal equipment. However, in trunk crossovers, cabling crossovers, and trunk wiring closets, crossover hardware using plug cables is usually already installed. In satellite wiring areas, such as wall-mounted cabling areas, crossovers may not require plug cables because the lines are often connected to I/O via jumpers. In the aforementioned locations in large cabling systems, crossovers often serve as transition points for converting large cables from the trunk subsystem to smaller horizontal cables connecting I/O. Feed-through crossovers are generally not used during cabling rerouting. Backbone Subsystem The backbone subsystem is part of the building's overall cabling system. It provides the routing for the building's backbone (feeder) cables. It typically provides multiple cabling facilities between two units, particularly at common system equipment located at a central point. This subsystem consists of all cabling cables, or a combination of conductors and fiber optic cables, along with the associated supporting hardware connecting these cables to other locations. Transmission media may include internal cabling vertically between floors in a multi-story building or cables from main units (such as computer rooms or equipment rooms and other backbone junction boxes). To communicate with other buildings within the building, the backbone subsystem connects trunk lines and cabling crossover points in equipment rooms to inter-building facilities to form a building cluster subsystem. To provide communication capabilities with external networks, the backbone subsystem connects trunk crossover points and network interfaces (part of the network facilities provided by the telephone exchange). Network interfaces are typically located in adjacent rooms of equipment rooms. Network interfaces define the boundaries between these facilities and the building's overall cabling system. Building Cluster Subsystem The building cluster subsystem extends cabling from one building to communication equipment and devices in other buildings within the building cluster. It is part of the overall cabling system (including the transmission medium) and supports the hardware required to provide communication facilities between buildings, including conductors, cables, fiber optic cables, and electrical protection devices to prevent surge voltages from entering the building. Equipment Cabling Subsystem The equipment cabling subsystem consists of cables, connectors, and related supporting hardware in equipment rooms, interconnecting various devices within a common system. This subsystem connects trunk crossovers and cabling crossovers to common system equipment (such as PBXs). This subsystem also includes wiring in equipment rooms and adjacent units (such as building entrance areas). These wires connect equipment or lightning protection devices to a valid building grounding point compliant with US Electrical Codes (NEC). Industrial Cabling System Industrial Ethernet cabling systems are designed for industrial automation objects such as FA (Factory Automation) and PA (Process Automation). While it shares some similarities with the cabling system defined by AVAYA, there are also significant differences. In its October 2003 "Planning and Installation Guide," IOANA recommended industrial cabling systems defined by EN 50173 and ISO/IEC 11801. Each building must have at least one building distributor (BD), and all BDs are connected to the campus distribution system (CD) via a star topology. The CD can be considered the central hub of the entire communication system, while redundancy may be used between BDs for security reasons. Within a building, the field-level cabling system (FD) can be deployed on different floors, each covering an area of ​​2000 square meters. Sometimes, due to the field distance being around 30 meters, the concept of the machine distribution system (MD) can be replaced by the FD. A comparison of the two diagrams shows that AVAYA's SYSTEM MAX definition of commercial networks is very detailed, with six subsystems comprehensively covering the physical layer cabling portion of commercial networks. The definitions of EN 50173 and ISO/IEC 11801 for industrial Ethernet physical layer cabling systems are more suitable for FA and PA. Currently, in the application of Industrial Ethernet in the industrial field, cabling systems often go directly from the Business Layout (BD) to the front-end devices. This is due to several reasons: 1. The horizontal cabling subsystem in automation is relatively simple compared to general cabling, requiring no patch panels, large-pair cables, or similar equipment. 2. SYSTEMAX's PDS system is a structured and modular cabling system designed to improve the flexibility and maintainability of the building's physical layer, which is not currently necessary in the FA/PA (Front-End/Package) field. 3. The absence of management subsystem equipment such as patch panels, patch cords, and wall mounts reduces intermediate steps, lowering the failure rate and making the system more reliable. However, with the increasing penetration of IP technology in the industrial field, if the field device layer also extensively adopts IP devices (such as sensor actuators), it is foreseeable that these two cabling diagrams will become increasingly similar, although the requirements for cabling equipment in industrial applications will be higher, and they will also support greater mobility. II. Topology Issues in Industrial Networks Topology is another issue to consider at the physical layer. Topology refers to how cabling is routed within the network. Point-to-point connections are connections between the interfaces of a workstation and a hub, between one hub and another, or between one workstation and another. Research on Industrial Ethernet shows that the topologies described in the EN50173 and ISO/IEC 11801 standards can be fully applied to industrial environments with minor modifications. Most users in industrial fields are more familiar with bus-type connections, where multiple workstations share a common connection. EIA-485 or Controller Area Networks (CAN) are good examples of these networks. However, bus topologies are no longer suitable for Industrial Ethernet. Although 10BASE2 and 10BASE5 are indeed bus-type coaxial cable-based Ethernet networks, their use is gradually decreasing due to their limitation to 10Mbps half-duplex operation and their exclusion from the emerging commercial building cabling standard TIA/EIA-568-A. For these reasons, Industrial Ethernet cabling often uses a star topology, requiring either connection hubs or switching hubs. Therefore, avoid using bus-based methods to connect transmission systems and similar networks, despite their simplicity. For industrial Ethernet, opt for star, tree, or ring topologies. In a typical industrial environment, we can divide the network into the following units: 1. CD == Campus distributor (industrial park level) 2. BD == Building distributor (factory level) 3. FD == Floor distributor (workshop level) 4. MD == Machine distributor (machine (equipment) level) 5. MO == Machine outlet (equipment output node) 6. TO == Terminal outlet (terminal output node) A typical industrial Ethernet topology is shown in Figure 2.3 below: [align=center]Figure 2.3 Industrial Ethernet Topology[/align] Each device output node connects to the network in a star configuration. This connection device is typically a switch. For example, Ethernet/IP, HSE, and EPA solutions all use a full-duplex switch + 100Base-TX topology. The advantage of a switch lies in its ability to eliminate frequent network collisions by changing the collision domain. However, collisions can still occur when two upstream ports attempt to send data to a downstream port simultaneously. Switches can effectively segment shared LANs, allowing each user to share maximum bandwidth. They can connect shared Ethernet segments and LANs of different speeds. Their switching technology operates at Layer 2 of the seven-layer network model, the data link layer, hence the term Layer 2 switching. Switches forward data packets based on Ethernet Destination Medium Access Control (MAC). By extracting the source MAC address of each data packet sent to the switch, the port obtains the MAC destination address and its relationship with the receiving port, thus determining the relationship between the port and the MAC destination address. Switches significantly improve the available bandwidth for users. However, because Layer 2 switching relies primarily on MAC addresses to transmit frame information, it uses continuously collected data to build an address table, recording the port from which each MAC address originates. Each Ethernet packet is then sent from the correct port. When a broadcast packet arrives, it must be forwarded to all ports of the switch, which can easily lead to broadcast storms in a network consisting only of switches. Furthermore, due to the continuous expansion of network scale, the combination of switches and routers is required, which has led to some shortcomings. This necessitates the development of Layer 3 and Layer 4 switching technologies, which will not be elaborated upon here. A typical device configuration using a switch connection is shown in the following figure: [align=center]Figure 2.4 A typical switch connection[/align] Another topology characteristic relevant to industrial environments is redundancy. In general commercial applications, Ethernet redundancy is not particularly important. Previously, hubs and switches were widely used to connect various Ethernet-based devices (such as PCs). A hub receives a message from one port and then broadcasts it to all other ports. For every message from any port, the hub forwards it to all other ports. A switch, however, can route messages from one port to another, automatically detecting the network speed of each network device. Using a feature called a "MAC address table," a switch can also identify and remember devices in the network. This intelligence avoids message collisions, improves transmission performance, and represents a significant improvement over hubs. However, devices like hubs and switches, while prioritizing ease of use and price, also lose the ability to implement advanced requirements such as redundancy. Subsequently developed managed switches, compared to hubs and ordinary switches, possess more complex functions and can typically be fully configured through network-based interfaces. They can automatically interact with network devices, and users can manually configure the network speed and flow control for each port. Most managed switches also offer advanced features such as SNMP (Simple Network Management Protocol) for remote monitoring and configuration, port mapping for diagnostics, VLANs (Virtual Local Area Networks) for grouping network devices, and priority sorting to ensure priority messages pass through. These new features enable the creation of redundant networks using managed switches. Using a ring topology, managed switches can form a ring network. Each managed switch can automatically determine the optimal transmission path and backup path, automatically blocking the backup path when the preferred path is interrupted. As the redundancy requirements of industrial networks have become more prominent, managed redundant switches with dedicated redundancy extensions have emerged. These switches offer special features, particularly optimized for redundant systems with stringent stability and security requirements. Common methods for building redundant networks include STP, RSTP, loop redundancy, and backbone truncating. 1. STP and RSTP: STP (Spanning Tree Protocol) exists as a link-layer protocol (IEEE 802.1D), providing path redundancy and preventing network loops. It forces backup data paths into a blocked state. If a path fails, the topology can be reconfigured and the link rebuilt by activating the backup path. Network outage recovery time is between 30-60 seconds. RSTP (Rapid Spanning Tree Algorithm, IEEE 802.1w), as an upgrade to STP, reduces network outage recovery time to 1-2 seconds. STP network structure is flexible, but its slow recovery speed makes it unsuitable for many industrial environments. 2. Loop Redundancy: Following STP, to meet the high real-time requirements of industrial control networks, loop connections were adopted to achieve fast redundancy recovery. This technology allows the network to recover automatically within 300ms after an outage. It can also alert users to network outages through faulty relay connections on switches, status indicator lights, and SNMP settings. These methods help diagnose where the ring network has broken down. Ring redundancy can generally be implemented using three methods: * Single-machine single-ring redundancy. * Dual-machine single-ring redundancy. * Dual-machine dual-ring redundancy. The main advantage of dual-ring redundancy is that it can avoid problems caused by a single cable failure through dual-channel connections. Dual-machine redundancy also avoids problems caused by a single device failure. 3. Trunk Redundancy Trunking Technology This method sets multiple ports on different switches as trunk backbone ports and establishes connections, forming a high-speed backbone link between switches. This not only significantly increases the network bandwidth of the backbone link and enhances network throughput, but also provides another function: redundancy. When a backbone link in the network experiences a disconnection, data in the network will be transmitted through the remaining links, ensuring normal network communication. The trunk backbone network can use bus or star network structures, and the theoretical communication distance can be extended indefinitely. This technology, employing hardware detection and data balancing, achieves a new level of network outage recovery time, typically below 10ms. Since redundancy significantly increases equipment costs, the choice of redundancy structure should be based on project requirements. Finally, a topology-related issue concerns cable length and connectivity. Commercial Ethernet follows the well-known 5-4-3 rule regarding cable length and connectivity: a network can have five segments connected by four repeaters, but only three segments can connect to computers. Similar rules still apply in industrial Ethernet environments. In fact, in industrial workshops such as chemical plants, cable length is a critical factor. A key consideration when deciding between twisted-pair cables and fiber optics is cable length. Without repeaters, cables are limited to 100m, while fiber optics can be limited to kilometers. Based on physical requirements, major industrial Ethernet solutions specify clear requirements for cable length and the number of connectable nodes. The relevant industry standard is ISO/IEC 11801. [15] The following is a typical workshop industrial network topology diagram: [align=center] Figure 2.6 A typical workshop industrial network topology[/align] In the above content, we have explained the relevant issues of the physical layer of industrial Ethernet in three main aspects. It should be said that as a reason for promoting Ethernet, low-cost commercial Ethernet equipment can be used in industrial control systems. In some applications, these applications are feasible. However, the factory environment has its own characteristics. Although everyone wants to use Ethernet chips and media that are available through commercial channels, the requirements of these factories for products are indeed different from those of offices. These reasons include general environmental factors, such as the effects of high temperature, humidity and vibration; physical environmental factors, such as EMC, radiation, etc.; and safety factors, such as power supply, grounding, etc. In industrial equipment that costs millions, strict equipment and environmental standards are indeed required. The following is a recommended standard of the Industrial Automation Open Networking Association (IAONA) for cables, connectors, etc. [15] [align=center] Table 2.4 A Recommended Standard for IAONA[/align] III. Issues Regarding the Working Mechanism of Front-End Devices For commercial Ethernet, gigabit or even 10-gigabit has become the choice for backbone networks, but it seems a bit ahead of its time for industrial Ethernet. After all, industrial networks require stability and real-time performance. Therefore, 100BASE-TX and full-duplex mechanisms have become the preferred choice for industrial networks. The standard bandwidth of 100M, which has been running stably in commercial networks for many years, can meet most of the communication requirements of FA/PA, and the full-duplex mechanism can overcome the network transmission uncertainty brought about by CSMA/CD. Currently, many PLCs, PACs, HMIs, and embedded devices on the market use the 10M half-duplex transmission mechanism. NIC front-end devices with this mechanism may not be fully compatible with the Layer 2 and Layer 3 network devices connected to them (electrical ports are 10/100M adaptively compatible, but optical ports are only 100M full-duplex incompatible), which may lead to uncertainties in data transmission. (Images of transceivers and optical switches provided by TSC Excellence Beijing) As shown in the diagram above, the transceivers and switches are cascaded via optical ports, both operating at 100M full-duplex, while the electrical ports are 10M/100M, with full-duplex/half-duplex auto-sensing. IV. Summary In conclusion, due to BELDEN's acquisition of HIRSCHMANN, a leading global provider of industrial network solutions, we are now focusing on the most easily overlooked layer in industrial networks—the physical layer. We have explored several issues: industrial Ethernet cabling, network topology, and network transmission mechanisms. A simple analysis leads to the following conclusions: 1. Currently, industrial Ethernet cabling systems are indeed in a primitive, rudimentary stage. With the increasing penetration of IP technology in the industrial field, and the widespread adoption of IP address addressing in field devices (such as sensors and transmission equipment), it is foreseeable that cabling systems and equipment used in the industrial field will change from the current "cable + connector" connection mode, becoming more complex, more flexible, with higher protection levels, and greater support for mobility. This presents significant business opportunities. 2. Without considering wireless network systems, the current topologies of industrial Ethernet are mainly star and ring topologies. For security and reliability considerations, redundant dual-star and redundant dual-ring topologies are the mainstream, while ring topologies seem to be more favored by automation users. 3. Due to the relatively small data volume but high real-time reliability requirements of industrial networks, 100BASE-TX and full-duplex should be the mainstream transmission mechanism for industrial network design now and for a considerable period to come. Adopting 10BASE-T and half-duplex transmission mechanisms will cause compatibility issues in the future.